Two novels were selected from two distinct periods in John Steinbeck's career, and text samples from each were pooled according to era. Using quantitative measures of text emotion, between-sample and norm-relative comparisons were computed to evaluate a priori expectations regarding differences in the use of implicit emotional information. These hypotheses reflected abstract characteristics that transcend the focus of a single novel. Analysis of word-based measures did not yield the between-sample differences expected. Rather, the hypotheses were supported by distinct patterns of phoneme distribution. The findings confirm and extend previous work by demonstrating that authors of prose manipulate the emotional quality of phonemes, and that the associated patterns can occur at a more abstract level than previously demonstrated.